In order to overcome the “semantic gap” and other problems faced by traditional video processing technologies, the information contained in the patio-temporal subspace of the 3D model is used for video processing and analysis, and an intelligent video surveillance technology guided by the spatiotemporal subspace of the 3D model is proposed. First, the 3D target model is matched from the video under the guidance of the body shape subspace. Secondly, we extract video events under the guidance of the patio-temporal subspace of the 3D model: surveillance object video + 3D model spatiotemporal subspace $\rightarrow 3\mathrm{D}$ action of the surveillance object. Finally, the action comparison classification in our 3D event library: motion data + 3D event library $\rightarrow$ video type and nature. Experimental simulation shows that the application of the three-dimensional intelligent panoramic monitoring system in the field of vehicle monitoring has achieved satisfactory results.